Skip to main content

Resultados relacionados

Resultados totales (Incluyendo duplicados): 3
Encontrada(s) 1 página(s)

e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

The spectra of irreducible matrices over completed idempotent semifields

  • Valverde Albacete, Francisco José
  • Peláez Moreno, Carmen
Motivated by some spectral results in the characterization of concept lattices we investigate the spectra of reducible matrices over complete idempotent semifields in the framework of naturally-ordered semirings, or dioids. We find non-none eigenvectors for every non-none element in the semifield and conclude that the notion of spectrum has to be refined to encompass that of the incomplete semifield case so as to include only those eigenvalues with eigenvectors that have finite coordinates. Considering special sets of eigenvectors brings out finite complete lattices in the picture and we contend that such structure may be more important than standard eigenspaces for matrices over completed idempotent semifields., FJVA was partially supported by EU FP7 project LiMoSINe (contract 288024) for this research. CPM was partially supported by the Spanish Government–Comisión Interministerial de Ciencia y Tecnología project 2011-26807/TEC.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

100% classification accuracy considered harmful: The normalized information transfer factor explains the accuracy paradox

  • Valverde Albacete, Francisco José
  • Peláez Moreno, Carmen
The most widely spread measure of performance, accuracy, suffers from a paradox: predictive models with a given level of accuracy may have greater predictive power than models with higher accuracy. Despite optimizing classification error rate, high accuracy models may fail to capture crucial information transfer in the classification task. We present evidence of this behavior by means of a combinatorial analysis where every possible contingency matrix of 2, 3 and 4 classes classifiers are depicted on the entropy triangle, a more reliable information-theoretic tool for classification assessment. Motivated by this, we develop from first principles a measure of classification performance that takes into consideration the information learned by classifiers. We are then able to obtain the entropy-modulated accuracy (EMA), a pessimistic estimate of the expected accuracy with the influence of the input distribution factored out, and the normalized information transfer factor (NIT), a measure of how efficient is the transmission of information from the input to the output set of classes. The EMA is a more natural measure of classification performance than accuracy when the heuristic to maximize is the transfer of information through the classifier instead of classification error count. The NIT factor measures the effectiveness of the learning process in classifiers and also makes it harder for them to "cheat" using techniques like specialization, while also promoting the interpretability of results. Their use is demonstrated in a mind reading task competition that aims at decoding the identity of a video stimulus based on magnetoencephalography recordings. We show how the EMA and the NIT factor reject rankings based in accuracy, choosing more meaningful and interpretable classifiers., Francisco José Valverde-Albacete has been partially supported by EU FP7 project LiMoSINe (contract 288024): www.limosine-project.eu Carmen Peláez Moreno has been partially supported by the Spanish Government-Comisión Interministerial de Ciencia y Tecnología project TEC2011–26807.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publication Artículo científico (JournalArticle)

The spectra of reducible matrices over complete commutative idempotent semifields and their spectral lattices

  • Valverde Albacete, Francisco José
  • Peláez Moreno, Carmen
Previous work has shown a relation between L-valued extensions of Formal Concept Analysis and the spectra of some matrices related to L-valued contexts. To clarify this relation, we investigated elsewhere the nature of the spectra of irreducible matrices over idempotent semifields in the framework of dioids, naturally ordered semirings, that encompass several of those extensions. This initial work already showed many differences with respect to their counterparts over incomplete idempotent semifields, in what concerns the definition of the spectrum and the eigenvectors. Considering special sets of eigenvectors also brought out complete lattices in the picture and we argue that such structure may be more important than standard eigenspace structure for matrices over completed idempotent semifields. In this paper, we complete that investigation in the sense that we consider the spectra of reducible matrices over completed idempotent semifields and dioids, giving, as a result, a constructive solution to the all-eigenvectors problem in this setting. This solution shows that the relation of complete lattices to eigenspaces is even tighter than suspected., FJVA was partially supported by EU FP7 project LiMoSINe (contract 288024) for this research. CPM has been partially supported by the Spanish Government Comisi on Interministerial de Ciencia y Tecnolog a project TEC2011-26807.